One of the benefits of cloud computing is the potential to reduce the per unit cost of computing services. Services are delivered on demand, so organizations no longer have to make massive upfront infrastructure investments that may end up being underutilized. However, this same benefit makes calculating those cost savings quite difficult. With traditional IT investments, the total cost is known (or at least estimated) up front. With cloud services, the aggregate cost can change over time as usage changes. Depending on the contract agreement, the per unit cost of the cloud service may decrease over time as well, leading to savings that cannot be anticipated at the outset. For example, the CIA expects that, similar to the commercial market, the service prices of its private Amazon- provided cloud service will decrease over time as adoption increases. Cloud services may fall prey to Jevon’s Paradox, which states that as technology becomes more efficient, the rate of use of that technology will increase, which leads to greater overall use and greater overall cost. (John Polimeni, 2007) As cloud computing becomes cheaper and the barriers to entry become fewer, the overall amount that the Department spends on it may rise, relative to what they might have spent on traditional IT.
Calculating cost savings or cost avoidance requires an understanding of the Total Cost of Ownership (TCO) of both the cloud service and the traditional, on-premises alternative. Savings are the difference between the TCO of the chosen solution and the TCO of the alternate solution. The majority of cloud costs are subscription costs, with subscriptions billed either per license or per unit of service cost (where unit of service is measured in computing time, number of requests processed, or megabytes stored or transferred). Cloud costs may also include implementation and training costs, especially if legacy applications are being migrated to the cloud. However, in some cases these aspects incur no additional costs if applications were originally designed for a virtualized or cloud environment.
An organization can measure traditional IT costs in capital expenditures, operating costs, and maintenance costs, which include labor costs to customize and maintain the hardware and software. Traditional IT costs include the cost of having unused IT capacity on standby as demand for an application or service grows. Traditional IT costs may also have long-term replacement costs once equipment reaches end of life (usually 5 years); those costs are invisibly integrated into cloud pricing.
Cloud computing costs are frequently more visible than on-premises costs, because the majority of cloud costs are wrapped into a known subscription fee. However, because cloud computing is billed as it is used (i.e., Pay As You Go), the aggregate amount paid for the subscription may change over time. As shown in Figure 3-1, the majority of on- premises IT costs are often hidden at the outset because they involve things like personnel and maintenance that may go beyond what is captured in a formal software license or IT management contract. Software licenses may also apply to cloud computing, depending on the type of cloud model being used (e.g., SaaS will include software licenses while IaaS may not).
Figure 3-1. Cost Comparison of On-Premises IT and Cloud Computing (Source: Mahoney, 2014)
In addition to the TCO of the system, moving from a traditional IT infrastructure to a commercial cloud infrastructure also introduces several unanticipated avenues of cost savings and expenditures. For example, by outsourcing servers from on-premises locations to the commercial cloud, military installations can achieve significant savings in
3-4
electricity costs. (Masanet, et al., 2013) In cases in which cloud implementations have greater availability and uptime than traditional implementations, cloud computing can lead to savings through increased productivity. However, as cloud use increases, there may be a need for increased bandwidth and an upgraded network to accommodate the larger bandwidth demands, driving up network costs. Architecting applications for the cloud environment can mitigate the larger bandwidth demands. Savings may also rely on shutting down and discontinuing support for legacy equipment, which the Government Accountability Office (GAO) found the Federal Government often has difficulty doing. (GAO, 2012) Without turning off legacy equipment and applications, an organization cannot achieve the anticipated cost savings from migrating to the cloud.
As DoD begins to collect data on cost savings, it must ensure that its IT contracting and financial management data is properly aligned to the cloud Pay As You Go model. Traditional IT is usually tracked as a capital expenditure (CAPEX), while cloud computing is tracked as an operational expenditure (OPEX). In DoD budgeting, separate money may be designated for CAPEX and OPEX, so the mission owners must ensure that they align their financial tracking to their cloud strategy. In a 2015 analysis, Gartner found that “few organizations have implemented financial management processes for public cloud, and therefore few have any idea whether they are saving money.” (Cancila, 2015) Gartner recommends that organizations define processes for continual financial monitoring, with appropriate alerts and triggers, so that cloud costs do not unexpectedly increase in the period between financial reviews. DoD may benefit from a centralized cost measurement function that can determine whether overall IT expenditures rise or fall as organizations adopt commercial cloud computing.